The term "honk" to me refers to the frequency
response.
When a graph is plotted that has frequency
on the X axis, and amplitude on the Y axis,
this "honk" could be clearly seen as a rise
or a peak in the plotted data.
Say, for instance, we plot frequencies from
0 to 20,000 cycles per second on the X axis.
Then, we can also plot amplitude in decibels
on the Y axis.
Without modern electronic equalization, we
will see many peaks and valleys in the plotted
frequency response.
Telephone circuits are famous for their
"honk" at 1000 cycles per second.
These resonant peaks are due to many factors.
Some few of them are: Playback horn shape.
Playback diaphragm resonance. Recording
horn shape. Recording diaphragm resonance.
There are many, many other factors too.
Think of the built-in resonance of the
reproducer's linkage, and then also consider
the resonant peaks built in to the hardware
used to carry the recorder cutter.
Most work on these factors was done on
an entirely empirical basis. The first attempt to throw some solid math at it, was
that Western Electric paper which resulted in
the Orthophonic machine.
So, they honk. But then again, so do our ears!
Chuck